We first cloned nisB into the pRSFDuet-1 vector using standard cloning techniques, digested with the restriction enzymes NdeI and XhoI. Gel electrophoresis, using highly purified low melting temperature agarose (BioWhittaker Molecular Applications, 50070), was used to isolate the desired fragments. An in agaro ligation generated a new plasmid. The ligation reaction mixture was directly transformed into E. coli DH5α cells for amplification. Plasmid was recovered by miniprep (Qiagen, 27106). Successful insertion was verified with restriction site mapping. This pRSFDuet-1-nisB plasmid became the foundation for five more plasmids, each containing a variation of nisA.

Alleles of nisA were inserted into pRSFDuet-1-nisB as described above, except restriction enzymes BamHI and HindIII were used for digestion. All five nisA alleles are approximately the same size; length ranges from 200 to 250 base pairs. Furthermore, they were all inserted into the same site on pRSFDuet-1-nisB. nisC was inserted into pACYCDuet-1 with restriction enzymes NdeI and XhoI. Figures 1 and 2 show the restriction site mapping used to verify insertion for the six plasmids necessary to our project: the five pRSFDuet1-nisA-nisB constructs and pACYCDuet-1-nisC. The backbone vectors pRSFDuet-1 and pACYCDuet-1 were used as negative controls.

Figure 1. Restriction digest verification of pACYCDuet-1-nisC. pACYCDuet-1 and pACYCDuet-1-nisC were digested with SspI. A 1% agarose gel with digested plasmids. B Plasmid maps with SspI cut sites are shown.

Figure 2. Restriction digest verification of five plasmids containing nisB and a nisA allele. A 1.3% agarose gel in which pRSFDuet-1 and each nisA plasmid were digested with BamHI and HindIII. Gamma = 0.4. Excised nisA bands are indicated with a red arrow. TL = truncated leader. B Plasmid maps with BamHI and HindIII cut sites are shown.

The experimental data matched our predictions for fragment size, suggesting successful plasmid creation. SspI cuts twice in pACYCDuet-1, generating fragments of 1727 and 2281 bp; SspI cuts three times pACYCDuet-1-nisC, generating fragments of 1405, 1727, and 2079 bp. Digesting with BamHI and HindIII generates fragments of ~200 and 6700 bp in pRSFDuet-1-nisA-nisB constructs and of 37 and 3792 bp in pRSFDuet-1. We further verified plasmid creation by outsourcing sequencing. We additionally made five more plasmids in DH5α strains by inserting the five nisA variants into pRSFDuet-1 directly, without nisB. We anticipated that removing nisB as another variable in protein expression might aid future troubleshooting.

Altogether, we made 31 different strains: 14 DH5α and 17 BL21(DE3), five of which were double transformants. Table 1 lists them by strain number and genotype. Transformation was verified in every strain by restriction site mapping at a minimum.

table of strains

Protein Expression

protein expression
Figure 3. NisA protein samples (induced with 0.4 mM IPTG at 37°C) compared to NisA, NisB, and NisC protein samples (induced with 0.4 mM IPTG at 18°C). All samples were expressed in E. coli BL21(DE3). A Coomassie-Blue stained 8-18% Tris-glycine SDS-PAGE. Bands of interest are marked with red asterisks. B Western blot on a nitrocellulose membrane with anti-His6 primary antibodies.

With our devices made, we next verified that the two-plasmid system was able to express NisA. Figure 3A shows an SDS-PAGE gel of induced cultures of cells with one of the nisA alleles and the nisB and nisC genes, alongside induced cultures of cells containing only the nisA alleles. The T7 promoter shows leaky expression when non-induced, so non-transformed E. coli BL21(DE3) cells serve as a negative control instead of non-induced cells with the genes. The predicted molecular weights of the five NisA variants are shown in Table 2. These proteins migrate anomalously slowly through an SDS-PAGE gel, appearing at a higher molecular weight (11-12 kDa) than their predicted molecular weights. These bands are visible on the corresponding Western blot (Figure 3). This anomalous migration of NisA is consistent with previous data from Chen and Kuipers (2021) [1].

table of weights

The predicted molecular weight of NisB is 118 kDa, while the predicted molecular weight of NisC is 48 kDa. Bands at this weight can be seen on an SDS-PAGE of protein samples of induced cultures of BL21(DE3) cells containing the nisA, nisB, and nisC genes (Figure 3A). Corresponding bands can be seen on a Western blot using anti-His6 antibodies for the same samples, along with a distinct band at 75 kDa and several low-intensity bands in between those three bands (Figure 3B). However, only the NisA variants have a His6 tag, suggesting some alteration to the migration of each of the NisA proteins through an SDS-PAGE gel is due to interactions between NisA and the post-translational modifiers NisB and NisC.

One possible explanation of the reduced mobility through an SDS-PAGE gel is the post-translational modifications themselves: NisC creates rings in the NisA primary structure through the five thioether bonds, creating a condensed secondary structure in NisA (see Experiments Figure 1D for active nisin structure). Another possible explanation is incomplete post-translational modification: during the enzymatic reactions catalyzed by NisB and NisC, transient covalent bonds are formed between NisA and the active site of each enzyme. Incomplete reactions could result in denaturation-stable NisA-NisB and NisA-NisC complexes that appear near the predicted molecular weights of NisB and NisC (Figure 3B). It is also possible that some NisA dissociates from NisB or NisC before the enzymatic reactions are complete. Eight serines and threonines are dehydrated by NisB and five thioether bonds are created by NisC; if less than eight of the residues are dehydrated or partially dehydrated, or if fewer than five thioether bonds are created, it would create variability in the migration of modified NisA through an SDS-PAGE gel. The distinct band that appears at 75 kDa may be NisA that has been fully modified by NisB and NisC, while the several bands of low intensity between the 48 kDa and 118 kDa bands may be partially modified NisA.

While the exact cause of the retarded migration of NisA through an SDS-PAGE gel in the presence of NisB and NisC is unknown, all five variants of NisA are affected similarly. NisB and NisC recognize the leader peptide of NisA, but it appears that the addition of TD-1, both on the N-terminus of the leader and between the leader and the nisin precursor, does not affect the specificity of NisB or NisC for NisA.


We first determined if NisA is a soluble or insoluble protein by fractionating IPTG-induced culture of BL21(DE3) cells containing a pRSFDuet-1-nisA plasmid. Figure 4 shows an SDS-PAGE gel with protein samples from each step of fractionation. We concluded that both soluble and insoluble forms of NisA exist from its presence in the pellet and supernatant. NisA being partially insoluble was further supported by the observation of inclusion bodies in the sonicate, under phase-contrast microscopy.

Figure 4. Monitoring effectiveness of cell fractionate with Coomassie Blue-stained 8-18% Tris-glycine SDS-PAGE gel. With BL21(DE3) containing pRSFDuet-1-nisA culture, expression was induced with 0.4 mM IPTG for 4 hours at 37°C (Lane 1). Half the culture was not induced (Lane 2). Cells harvested from induced culture were sonicated (Lane 3) and then centrifuged at 13,000g. Protein samples were made of medium-speed supernatant (Lane 4) and medium-speed pellet (Lane 5). NisA bands are labeled with a red arrow.

Because NisA is partially soluble, we moved forward with immobilized metal affinity chromatography to purify unmodified soluble NisA. To prevent column clogging, the medium speed supernatant was centrifuged at 100,000g and the high-speed supernatant collected. High-speed supernatant was applied to nickel-nitrilotriacetic acid beads and fractions were collected from the flow through, wash, and three elution fractions.

Figure 5. Coomassie Blue-stained 8-18% Tris-glycine SDS-PAGE gel of elution fractions after purifying unmodified NisA from high-speed supernatant of fractionation on Nickel-NTA column. Bands of interest are marked with a red asterisk. MW of ladder bands have been labeled in kD.

Figure 5 shows an SDS-PAGE gel of protein samples made from each fraction. The solitary bands in the Elution Buffer 2 fractions are of the predicted molecular weight of unmodified NisA. We pooled the eluate fractions containing pure unmodified NisA and characterized them with a BCA Pierce Kit at 20 µg/mL.

We have not yet purified the modified NisA proteins from a BL21(DE3) strain containing the nisA, nisB, and nisC genes. Early attempts showed that the protocols for induction and fractionation needed to be modified to accommodate the metabolic burden of NisB and NisC being produced alongside NisA.

Activity and Permeation

Once modified NisA is produced, we need a method to verify its antimicrobial properties. A disk diffusion assay was adapted for this need, but first this assay had to be optimized for nisin. We found that E. coli was an inadequate positive control strain for nisin because the antibiotic is only active against Gram-positive bacteria. Thus, we changed to M. luteus for plating. M. luteus does not grow optimally on LB agar, so we replaced the LB agar plates with M16 + 5% glucose plates. We found that nisin poorly diffuses through agar; zones of inhibition were significantly smaller for nisin than kanamycin at equivalent concentrations (Figure 6). To increase the limit of detection of active nisin for our assay, we supplemented our nisin stock with 0.01% Tween 20. We posited that the non-ionic detergent would better solubilize the amphipathic nisin in the agar. Figure 6 shows the culmination of these efforts. Adding Tween 20 greatly increases nisin’s zone of inhibition without demonstrating antimicrobial activity itself.

disk diffusion
Figure 6. Disk diffusion asay of nisin on M16 + 5% glucose agar plate covered in M. luteus. Disks were loaded with 10 µL solution. Kanamycin at 5 mg/mL was used as a positive control. Nisin was dissolved at 5 mg/mL in a Sodium Phosphate Buffer pH 6.0, used as a negative control (Buffer). Tween 20 was added to labeled samples to final concentration 0.01% v/v. Zones of inhibition are labeled in cm.

When we purify modified NisA variants, we will add Tween 20 to raise the detection threshold for antimicrobial activity. We also plan to change the strain used in plating from M. luteus to S. aureus to better model our device’s efficacy.

To test permeation, we utilized the Franz diffusion assay. Before using the assay with nisin, its reliability must be tested with a positive control to establish a functional assay for the experimental variants. Caffeine was chosen as a positive control based on literature [2]. Caffeine absorbs light with a peak at 274 nm, so we first plotted the absorbance of caffeine at 274 nm against its concentration to calculate an extinction coefficient. Figure 7 shows that our line of best fit for this data demonstrates strong linearity.

caffeine absorbance
Figure 7. Calculating the extinction coefficient of caffeine. Concentrations were prepared in triplicate. The error bars shown represent one standard deviation from the mean. Absorbances were measured with a NanoDrop Spectrophotometer.

Next, the permeation of caffeine across skin was measured. Figure 8 shows that caffeine did accumulate over time in the receptor compartment, suggesting permeation through the skin. After 6 hours, 74.4% of the caffeine had diffused through the skin. We applied a linear fit to the curve according to diffusion equations generated with modeling; the data showed a strong fit. Our next step will be to repeat this experiment to validate the results.

caffine over time
Figure 8. Cumulative mass of permeated caffeine over time. 1 mL of caffeine at 250 µg/mL in PBS was loaded into the donor compartment of the Franz diffusion cell. 20 µL samples of the receptor compartment were taken at 1.5, 2, 4, and 6 hours. The absorbances of the samples were measured and the linear model was used to calculate the caffeine concentration of the samples. To calculate total permeated caffeine, the concentration was multiplied by the total volume of the receptor compartment, 9 mL.

Once we have purified modified NisA from our device, we will use this functional assay to measure permeation. To measure the concentration of NisA in the receptor compartment, we will switch from a spectrophotometer to a BCA Pierce Protein Concentration Kit. The four relevant experimentals are V8-NisA, TL-TD1-V8-NisA, TD1-V8-NisA, and TD1-linker-V8-NisA. We predict that the addition of TD-1 will increase permeation through the skin.


  1. Chen, J., & Kuipers, O. P. (2021). Isolation and Analysis of the Nisin Biosynthesis Complex NisBTC: Further Insights into Their Cooperative Action. mBio, 12(5), e02585-21.
  2. van de Sandt, J. J., van Burgsteden, J. A., Cage, S., Carmichael, P. L., Dick, I., Kenyon, S., Korinth, G., Larese, F., Limasset, J. C., Maas, W. J., Montomoli, L., Nielsen, J. B., Payan, J. P., Robinson, E., Sartorelli, P., Schaller, K. H., Wilkinson, S. C., & Williams, F. M. (2004). In vitro predictions of skin absorption of caffeine, testosterone, and benzoic acid: a multi-centre comparison study. Regulatory toxicology and pharmacology : RTP, 39(3), 271–281.